AT domain classification on basis of the NJ-algorithm for various selected sets of residues.

The first column lists the different substrate groups and gives the number of represented sequences between brackets. The values in columns 3, 4 and 5 were calculated on basis of the residues identified by [24], [25] and [35], as indicated. The two major substrate groups MC (malonyl-CoA) and MMC (methylmalonyl-CoA) were reasonably well distinguishable in all trees. However, the factual accuracy of the MC and MMC prediction is lower than 1 as all of the ‘minor’ substrate specific AT sequences fall within the both clusters. Abbreviation: nsc, not in a single cluster.$ For substrate abbreviations see the legend of Figure 2. The initial complete dataset was used to compose the Table (i.e. including the near duplicate sequences), excluding the sequences related to BzC (2), 3MbuC (1), AC (1), CH (1), and CP (1).a* 165 conserved positions (100% identity) in at least one of the substrate groups; 146 conserved positions in case the residues are removed that are conserved throughout all substrate groups; b* Conserved positions (100% identity) in at least three of the substrate groups (do not include global identical).